The proposed 5-year continuation research project will assess longitudinal outcomes during the transition from adolescence into early adulthood for a large 2-group sample of 401 at-risk urban youth: (1) a probability sample of 251 homeless adolescent from throughout the Detroit metropolitan area and (2) 150 initially housed adolescents matched on gender, age, race, and SES. Measures assessed longitudinally include family and contextual factors, psychological processes (e.g., competence, self-efficacy), homelessness and stress, peer relations, and outcomes that often begin in late adolescence and coalesce in early adulthood (e.g., alcohol and drug use/abuse, criminal behavior, education and employment, mental and physical health, HIV-risk behaviors). The proposed research builds on the original study by extending the follow-up period from adolescence (initial-final mean ages= 15.5-17.0) into early adulthood (initial-final mean ages=19.5- 21.5). We believe that is during this period of early adulthood that the consequences of problematic experiences in adolescence will become most clearly visible. Relatively few studies have tracked at-risk youth through the important transition. Specifically we will: (1) investigate risk and protective factors predicting longitudinal outcomes and test theoretical models on longitudinal pathways in our at-risk sample; (2) improve understanding of family dynamics among at-risk urban youth during the transition to early adulthood; (3) assess the impact of peer relationships; and (4) assess longitudinal differences in outcomes for African-vs. European-Americans and for girls vs. boys. In the original 4- year project, the rate of attrition has been low and we expect that 179 (81%) of the initially homeless and 141 (94%) of the housed ill receive at least one follow-up interview during early adulthood in the continuation project. Preliminary findings over 1.5 years suggest that our at-risk youth show some short-term improvements (perhaps due to service, family, or other factors) but are beginning to show signs of negative longitudinal outcomes (e.g., alcohol and drug problems) as they turn 16-17. W e are also finding some support for out theoretical perspectives. For example, supporting a social learning perspective, we have found that family relations, school problems, and deviant peers all have roles in the development of alcohol use/abuse.